Fibre Channel (SAN)

Reply
Occasional Contributor
Posts: 8
Registered: ‎12-18-2012

FCIP and TI Zones - Failure in an FCIP circuit did not lead to Failover

We have an FCIP configuration between two 7800 switches.

We do not currently have the Advanced Extension License, so are using two FCIP Circuits and manage the availability via Traffic Isolation zones.

We experienced a brief failure in one of the FCIP circuits the other day. However, it lasted less than 3 seconds. During this time, the traffic did not failover to a non-dedicated path.

As such, the failure was seen by the VPLEX Disk arrays which use the FCIP link:


>> (A2-FC01): In the past minute, 2.20% percent IO failed on I 0x0x500014426049c821FS64x" T 0x0x5000144260614521FS64x" (nportid 0x0e0200) due to IO timeout, there might be faulty hardware in the IO path.
>> Director{2.1.40.84.0}] RCA: IO failed due to exchanges timed out. There might be faulty hardware on the IO path.
>> IO exchange timeout threshold exceeded.

My questions are, at what point will the Traffic Isolation Zone failover to the non-dedicated path? Is there a timeout parameter? Is it tuneable?

The configuration details are quite simple and are as follows:

Kernel:     2.6.14.2

Fabric OS:  v6.4.2a

Made on:    Mon Jul 18 22:35:02 2011

Flash:      Sat Apr 6 23:35:08 2013

BootProm:   1.0.9

A switchshow output details:

hl-fcsw-slo-03:admin> switchshow
switchName:     hl-fcsw-slo-03
switchType:     83.3
switchState:    Online
switchMode:     Native
switchRole:     Principal
switchDomain:   3
switchId:       fffc03
switchWwn:      10:00:00:05:33:54:a3:7a
zoning:         ON (ZS_STRETCHED_FABRIC_C)
switchBeacon:   OFF
FC Router:      ON
FC Router BB Fabric ID: 1
Address Mode:   0

Index Port Address Media Speed State     Proto
==============================================
  0   0   030000   id    N8   No_Light    FC  Disabled (Persistent)
  1   1   030100   id    N8   No_Light    FC  Disabled (Persistent)
  2   2   030200   id    N8   Online      FC  F-Port  50:00:14:42:60:49:c8:20
  3   3   030300   id    N8   Online      FC  F-Port  50:00:14:42:70:49:c8:20
  4   4   030400   --    N8   No_Module   FC  Disabled (Persistent)
  5   5   030500   --    N8   No_Module   FC  Disabled (Persistent)
  6   6   030600   --    N8   No_Module   FC  Disabled (Persistent)
  7   7   030700   --    N8   No_Module   FC  Disabled (Persistent)
  8   8   030800   --    N8   No_Module   FC  Disabled (Persistent)
  9   9   030900   --    N8   No_Module   FC  Disabled (Persistent)
10  10   030a00   --    N8   No_Module   FC  Disabled (Persistent)
11  11   030b00   --    N8   No_Module   FC  Disabled (Persistent)
12  12   030c00   --    N8   No_Module   FC  Disabled (Persistent)
13  13   030d00   --    N8   No_Module   FC  Disabled (Persistent)
14  14   030e00   --    N8   No_Module   FC  Disabled (Persistent)
15  15   030f00   --    N8   No_Module   FC  Disabled (Persistent)
16  16   031000   --    --   Online      VE  VE-Port  10:00:00:05:33:69:3d:d4 "hl-fcsw-lhr-03"
17  17   031100   --    --   Online      VE  VE-Port  10:00:00:05:33:69:3d:d4 "hl-fcsw-lhr-03" (downstream)
18  18   031200   --    --   Offline     VE
19  19   031300   --    --   Offline     VE
20  20   031400   --    --   Offline     VE
21  21   031500   --    --   Offline     VE
22  22   031600   --    --   Offline     VE
23  23   031700   --    --   Offline     VE
     ge0  cu    1G   Online    FCIP  Copper
     ge1  cu    1G   Online    FCIP  Copper
     ge2  --    1G   No_Module FCIP  Disabled
     ge3  --    1G   No_Module FCIP  Disabled
     ge4  --    1G   No_Module FCIP  Disabled
     ge5  --    1G   No_Module FCIP  Disabled


As recommended by EMC/Brocade were using TI Zoning:

hl-fcsw-slo-03:admin> zone --show
Defined TI zone configuration:

TI Zone Name:   VPLEX_WAN_0

Port List:      13,2; 13,3; 13,16; 3,16; 3,2; 3,3

Configured Status: Activated / Failover-Enabled
Enabled Status: Activated / Failover-Enabled

In an active/failover-enabled configuration.

On the FCIP Circuit side, we have:

hl-fcsw-slo-03:admin> portshow fcipcircuit all
-------------------------------------------------------------------------------
Tunnel Circuit  OpStatus Flags   Uptime   TxMBps   RxMBps ConnCnt CommRt  Met
-------------------------------------------------------------------------------
16     0 ge0     Up      ----s   1d9h9m     1.29     0.20    2    40/40    0
17     0 ge1     Up      ----s   57d23h     0.00     0.00    1    40/40    0
-------------------------------------------------------------------------------
  Flags: circuit: s=sack


hl-fcsw-slo-03:admin> fabricshow
Switch ID   Worldwide Name           Enet IP Addr    FC IP Addr      Name
-------------------------------------------------------------------------
  3: fffc03 10:00:00:05:33:54:a3:7a 172.32.90.54    0.0.0.0        >"hl-fcsw-slo-03"
13: fffc0d 10:00:00:05:33:69:3d:d4 172.31.90.54    0.0.0.0         "hl-fcsw-lhr-03"


Brief failure in Tunnel 16:

2013/06/02-23:42:51, , 68, CHASSIS, INFO, MP_7800B, FCIP Tunnel 16 High-Pri QoS UP.
2013/06/02-23:42:48, , 67, CHASSIS, INFO, MP_7800B, FCIP Tunnel 16 Low-Pri QoS UP.
2013/06/02-23:42:48, , 66, CHASSIS, INFO, MP_7800B, FCIP Tunnel 16 Med-Pri QoS UP.
2013/06/02-23:42:48, , 65, CHASSIS, INFO, MP_7800B, FCIP Tunnel 16 UP.
2013/06/02-23:42:48, , 64, CHASSIS, INFO, MP_7800B, FCIP Tunnel 16 Circuit 0 UP.
2013/06/02-23:42:45, , 63, CHASSIS, ERROR, MP_7800B, FCIP Tunnel 16 DOWN (Network/Remote/Other).
2013/06/02-23:42:45, , 62, CHASSIS, ERROR, MP_7800B, FCIP Tunnel 16 Low-Pri QoS DOWN (Internal Close).
2013/06/02-23:42:45, , 61, CHASSIS, ERROR, MP_7800B, FCIP Tunnel 16 Circuit 0 DOWN (Keepalive Timeout).
2013/06/02-23:42:45, , 60, CHASSIS, ERROR, MP_7800B, FCIP Tunnel 16 Med-Pri QoS DOWN (Keepalive Timeout).
2013/06/02-23:42:45, , 59, CHASSIS, ERROR, MP_7800B, FCIP Tunnel 16 High-Pri QoS DOWN (Keepalive Timeout).

Since, however, circuit 17 was unaffected – I’m curious as to why it wasn’t used? i.e. Why didn’t the TI zone failover to use a non-dedicated path?

Furthermore, if we go down the route of spending a load of money to get the Advanced Extension license in order to roll the two FCIP circuits into a single tunnel (and therefore dispense with the TI Zones), would this issue be masked from the disk array in future?

Regular Contributor
Posts: 178
Registered: ‎04-21-2008

Re: FCIP and TI Zones - Failure in an FCIP circuit did not lead to Failover

Hi grant,

assuming you have no other traffic than those vplex relationships, I would try to create 2 TI zones with failover disabled. Then you would two dedicated paths :

13,2; 13,16; 3,16; 3,2

and

13,3; 13,17; 3,17; 3,3

You would use both FCIP circuits all the time, and in case of failure on one of them VPLEX multipathing should be able to keep service available on one path. No more issue with TI zone failover.

BTW if you have money, I can tell you that FCIP trunking works very fine

Hope this helps

--

david

Occasional Contributor
Posts: 8
Registered: ‎12-18-2012

Re: FCIP and TI Zones - Failure in an FCIP circuit did not lead to Failover

David,

Many thanks for your response. I will investigate your suggestion with EMC. I believe without the AE License what we implemented was as per EMC's recommendation. I have a feeling (although I don't have the documentation at my finger tips) that the Version of Firmware we're running on the VPLEX arrays limited us to this configuration. I seem to recall that something to do with Fast Write limited us to shunting traffic down only one tunnel at a time.

I have recommended the AE License to avoid these issues moving forwards - so hopefully that will get taken up !

Once again, many thanks.

Best regards,

Grant

Join the Community

Get quick and easy access to valuable resource designed to help you manage your Brocade Network.

vADC is now Pulse Secure
Download FREE NVMe eBook